What hurts the dominant airlines at hub airports? Barcelona, May 25th 2015
Xavier Fageda Associate Professor (
[email protected])
Introduction
• But……….in Europe, this is changing (examples; Difficulties of Iberia & Alitalia, KLM -Transavia, Bankrupcy of Malev, Ryanair and Norwegian in main airports….) • In 2002-2013, network airlines’ share has fallen in 17 of 22 large European airports that have traditionally been dominated by former flag carriers. • A loss in the competitiveness of the dominant network airlines More traffic, more intercontinental destinations: Impact on firm’s location choices Airport dominance and high route concentration may lead to higher average air fares. Also, downward pricing pressure of LCCS on routes they operate
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
• While inter-hub competition may be intense, at their own hubs network airlines have typically benefited from weak competition with LCCs.
2
Introduction • Aims But……….to determine which dimensions of competition might undermine the competitive position of dominant airlines at hub airports • I identify the competitive position of each airline by the flight frequencies they are able to provide on a given route. • Hence, I estimate an equation in which the dependent variable is the frequencies offered by European network airlines on routes departing from their hub airports, using data for the period 2002-2013. • Questions Do network airlines reduce frequencies at their hubs when they are competing with low-cost airlines? Does competition takes place at the route, airport and/or city-
pair levels?
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
NOT to examine welfare implications of LCCs entry in hub airports
3
Contributions • Literature on airline frequency competition at the route level (Schipper et al.,
Focus on route competition (HHI_route, DLCC, Dhub) • Literature on the impact of LCCs on capacity competition (Goolsbee and Syverson, 2008; Bettini and Oliveira, 2008)
Dummies for multi-airport cities, presence of LCCs in both
airports • Literature on the effects of mergers on the distribution of traffic between airports (Bilotkach, Fageda & Flores-Fillol., 2013) Merger Delta & Northwest I add to this literature by including the following variables in the frequency equation: HHI (route, airport), Share_LCCs (route, airport), Dmerger, Frequenciessecondary_airport An important difference with previous studies is the focus on choices of network airlines at their hub airports.
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
2002; Richard, 2003; Pai, 2010; Bilotkach, Fageda & Flores-Fillol., 2010; Brueckner and Luo, 2013)
4
Sample
• I include 22 large EU airports at which the same airline was dominant throughout the period of study and at which that dominant airline was not a low-cost carrier. Air France (Paris-CDG, Paris-Orly), Air Lingus (Dublin), Alitalia (Rome-FCO), Austrian Airlines (Vienna), British Airways (London-LHR, London-LGW), Czech Airlines (Prague), Iberia (Madrid), Finnair (Helsinki), KLM (Amsterdam), LOT (Warsaw), Lufthansa (Frankfurt, Munich, Dusseldorf), SAS (Stockholm-ARN, Copenhagen, Oslo-OSL), SN Brussels (Brussels), Swiss (Zurich), TAP (Lisbon) and Tarom (Bucharest).
• A number of large EU airports are not included because we were unable to identify one dominant airline operating out of them for the whole period (BUD, MXP, ATH, MAN, BCN) or because they are dominated by LCCs (PMI, TXL, LTN, STN) • The exclusion of important European airports from our sample is a limitation but our restricted sample covers most of hubbing operations in EU. • I have complete data for 936 routes and the sample contains 11,232 observations.
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
• Route-level data from large EU airports to European and non-European destinations from 2002 to 2013.
5
Descriptive data
• Mean share of network airlines > 50% with some few exceptions. It has been substantially reduced in 17 of the 22 airports. This is quite remarkable if we consider that our sample of airports excludes those that are the home base of LCCs • Evolution in total traffic is quite diverse with some airports recording substantial growth (HEL, LIS, OSL, OTP) and others recording losses (ARN, BRU, MAD). • Only reduction in total number of non-stop destinations in LHR, FRA and ZRH (no data analyzed on intercontinental flights) but dominant airlines have reduced the number of non-stop destinations in 8 of the 22 airports. • Overall, it is not clear the aggregated effect of the increasing presence of LCCs airlines at European hub airports but they may have weakened the dominance of the former flag carriers.
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
• Mean Traffic share of the dominant airline always > 30% and in some cases > 60%. It has been reduced in 14 of 22 airports.
6
• Network airlines: Former flag carriers and/or airlines integrated in alliances • LCCs: Air Arabia Maroc, Air Finland, Air One, Air Transat, Air Berlin, Alpi Eagles, Blue Air, bmi baby, Cai Second, Centralwings, Condor, Corsairlfy, dba, easyjet, Fly Me, FlyNordic, Germanwings, Hapagfly, Jet2, Jet4you, LTU, Monarch, MyAir, MyTravelLite, NIKI, Norwegian, OLT, Ryanair, SkyEurope, Smart Wings, Sterling, Thomson, Transavia, TUIFly, Virgin Express, Volare, Vueling, Wind Jet, Wizzair, Zoom Air • Currently some hybrid LCCs (fare bundling and/or connecting flights) but very recent trend: Air Berlin, Germanwings, Norwegian, Vueling
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
Network vs Low cost airlines
7
Data limitations
Aggregated supply data; I cannot distinguish between direct and connecting passengers Results of the analysis should be complemented with studies that use origin and destination data of passengers. • Some potential omitted factors: Airport congestion (airport and year dummies may help) • No data on prices
Airline behavior can be viewed as a multistage decision process: 1) Enter in the market or not; 2) Once entered, routes and aircraft technology, 3) Flight schedule, 4) Prices (most flexible variable). Focus on previous (and this) study is on 3) Similar explanatory variables for price and frequency equations: impact of pricing strategic decisions of airlines can be indirectly considered through competition variables.
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
• Under hub-and-spoke structures, the service levels of airlines in the route will depend on the amount of traffic related with direct and connecting passengers
8
Variable
Description
Expected relationship
Frequenciesdominant_airline
Total number of annual flights offered by the dominant airline
Dependent variable
Populationdestination
NUTS 3 (Europe), Metropolitan areas (Rest)
Incomedestination
NUTS 3 (Europe), Index_country (Rest)
Demand (+)
Demand (+)
Distance
Number of kms flown
Intermodal competition, larger planes (-)
DEU
Dummy for intra-EU routes
No regulatory restrictions, stronger integration (+)
DUS_openskies
Dummy for US destinations after the OS agreement
No regulatory restrictions (+) More competition (-)
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
Variables
9
Variables Description
Expected relationship
Dinterhub_same_alliance
Dummy for routes connecting two hubs of airlines integrated in the same alliance
Proxy for code-share agreements (-)
Hub_competition
Number of airports in the sample that have direct flights to the destination
Demand (+), Inter-hub competition (-)
Droutes_with_multiple origin_airports
Dummy for routes in which air services are offered simultaneously from two origin airports (LHR, LGW; CDG, ORY).
Demand (+)
Dmerger
Dummy variable for routes in which the dominant airline was acquired by another larger airline.
Reorganization of route network in favor of airports of larger airline (-)
Dorigin_airport
Dummies for the airports of origin
Control for time-invariant airport-specific omitted variables
Dyear
Dummies for years
Control for the common trend on all routes
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
Variable
10
Variables Variable
Description
Expected relationship
Herfindahl-Hirschman index in - A possible response of an terms of flight frequencies, incumbent : Cut fares and add Share of low-cost airlines flights to boost demand - Less demand from direct passengers
HHIorigin_airport(+), Share_lowcostorigin_airport (-)
Herfindahl-Hirschman index in Less frequencies in a route may terms of flight frequencies, affect other routes: Share of low-cost airlines Less demand from connecting passengers due increased connection times
Frequenciessecondary_airport Number of flights offered by - More demand (+) LCCS in the same city-pair market from a nearby secondary - Competition (-) airport
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
HHIroute (+), Share_lowcostroute (-)
11
Econometric issues
• Models to consider unobserved route heterogeneity: Route random vs fixed effects FE: Control for omitted variables that correlate with variables of interest and do not change over time but….effect of time-invariant variables cannot be identified and imprecise estimations when low within-variation RE: Captures both between and within variation while FE only captures within variation but………. potential bias from correlation between explanatory variables and RE. Hausman test not useful when within variation low (Clark and Linzer, 2012; Troeger, 2008) • Lagged dependent variable as a regressor? Annual versus quarterly data, low withinvariation • Potential endogeneity of route competition variables: use of lags
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
• Breusch-Pagan / Cook-Weisberg test for heteroskedasticity: PROBLEM •Wooldridge test for autocorrelation in panel data: PROBLEM • Levinlin panel data unit root test: NO PROBLEM
12
Results (RE)
Explanatory variables
All (1)
Intra-EU (2)
Populationdestination
0.028***
Incomedestination
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
Dependent variable: Frequencies of dominant airline
0.10***
Intra-EU < 900 kms (3) 0.12***
Intra-EU routes > 900 kms (4) 0.10***
239.82***
7.09***
11.55***
6.94***
Distance
-0.14***
-0.89***
-1.03***
-0.47***
DEU
376.54***
-
-
-
DUS_openskies
-1.68
-
-
-
Dinterhub_same_alliance
-24.05
-15.53
-32.79
-0.07
Hub_competition
4.18**
5.27**
5.27
9.07***
Droutes_with_multiple origin_airports
3462.99***
3451.57***
3766.04***
1049.10***
Dmerger
-19.77*
-29.28**
-32.53
-20.85
13
Results (RE)
Explanatory variables
All (1)
HHIroute
18.14
Intra-EU (2) 15.32
Intra-EU < 900 Intra-EU routes kms (3) > 900 kms (4) -14.94 48.10**
Share_low-costroute
-162.33***
-196.19***
-313.15***
-36.39
HHIorigin_airport
294.48***
422.95***
492.96***
297.39***
Share_low-costorigin_airport
-206.36***
-205.47**
-533.54***
145.49
Frequenciessecondary_airport
0.04
0.03
0.13***
-0.04
Intercept
-2.20
1442.1***
1143.40***
1155.89***
R2
0.44
0.46
0.45
0.60
N
10296
6985
4114
2871
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
Dependent variable: Frequencies of dominant airline
14
Results (FE) • Stronger statistical significance of the merger variable Share_lowcostroute,
Share_low_costoriginairport,
• No statistical effect of HHIairport in regressions for all sample and long-haul intra-EU route •R2: 0.04
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
• Same results for HHIroute, Frequenciessecondary_airport
15
Caveats on results
In terms of elasticities: a 10% increase in the share of LCCs in the airport implies about a 2% decrease in frequencies, while a 10% increase in the share of LCCs in the route implies about a 1% decrease in frequencies.
An airline needs to provide 18422 annual flights in the airport (355 weekly flights) to achieve a 10% increase in that airport while an airline needs to provide 175 annual flights in the route (3 weekly flights) to achieve such increase in the route. • Frequencies offered at secondary airports are too small to impact the dominant carrier. LCCs in secondary airports provide an average frequency of two flights per week compared to the dominant carrier frequency of 4 flights per day in the primary airport.
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
• It seems that the impact of the share of low-cost carriers in the airport is higher than the impact of their share in the route………………
16
Conclusions
Negative effects suffered by dominant airlines are not only felt on the routes on which they compete directly, but also on other routes that may suffer a reduction in demand from connecting passengers
• Implications for airport management: Capacity expansions at large airports are subject to restrictions. A reasonable objective of airport managers should be achieving the best possible use of the current capacity by airlines operating in hubs. An increase of airport charges may have a differential impact on network and low-cost airlines. The introduction of market-based mechanisms in the allocation of slots could ensure that they go to the airlines able to make the best use of them.
GiM - Departament de Política Econòmica | Facultat d’Economia i Empresa Av. Diagonal, 690 08034 Barcelona | T. (+34) 93 402 19 43 | www.ub.edu/gim
• Dominant airlines may be worried by the increased presence of low-cost airlines at their hub airports
17
